PubMed Health⌕ Search

SEARCH · PubMed Health

Results for “predictive modelling”

Explore indexed PubMed citations for clinical trials, systematic reviews and public health research. Read source abstracts and follow each citation to its original PubMed record.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 1,639 records · Page 91Linked to original sources
PubMed2006

The utility of extended longitudinal profiles in predicting future health care expenditures.

BACKGROUND: Health care spending is highly concentrated. Prediction models that accurately identify the characteristics of individuals most likely to incur high levels of health expenditures in a subsequent year are important analytical and statistical tools. OBJECTIVES: This study examined the capacity of alternative models to predict the likelihood of incurring high levels of medical expenditures in a subsequent year. This effort also evaluated the utility of an additional year of longitudinal information. SUBJECTS: A nationally representative sample from the Medical Expenditure Panel Survey (MEPS). METHODS: The MEPS longitudinal data are used to examine the persistence of high expenditures during a 2-year period. With the unique linkage of the MEPS to the National Health Interview Survey, the utility of an additional year of data also was examined. Resultant models were evaluated in terms of sensitivity, specificity, and predictive capacity. RESULTS: Only modest marginal gains in discrimination capacity were realized from the use of extended longitudinal profiles from the National Health Interview Survey, relative to information on prior year characteristics. CONCLUSIONS: Our results highlight the continuing concentration of health care expenditures during the period 1996 to 2002 and reveal some attenuation in magnitude in the tail of this distribution over time. Further, our results provide evidence of the utility of probabilistic models as prediction tools to identify individuals likely to incur high levels of expenditures in future years. Predictive capacity does not suffer when restricted to a single year of prior information.

Adolescent↗
PubMed2026

Catecholaminergic polymorphic ventricular tachycardia mediated by ryanodine receptor 2: a validated risk stratification.

BACKGROUND AND AIMS: Patients with catecholaminergic polymorphic ventricular tachycardia (CPVT) are at risk for potentially life-threatening arrhythmic events (AEs) even while treated with β-blockers. The aim was to develop a model for individualized prediction of AEs in patients with RYR2-mediated CPVT on β-blocker monotherapy. METHODS: The derivation and independent validation cohorts included 743 and 129 patients, respectively. AEs were defined as arrhythmic syncope, appropriate implantable cardioverter-defibrillator shock, sudden cardiac arrest (SCA), and sudden cardiac death. Near-fatal or fatal AEs (nf/fAEs) included all AEs except for arrhythmic syncope. Prediction models using Cox regression were developed and internally and externally validated. RESULTS: A total of 102 (13.7%) patients in the derivation cohort and 24 (18.6%) patients in the validation cohort experienced ≥1 AE over a median follow-up of 5.1 [interquartile range (IQR), 7.7] and 2.4 (IQR, 4.4) years, respectively. Predictors of AE were arrhythmic syncope or SCA prior to diagnosis and age at β-blocker initiation. In the derivation and validation cohorts, the optimism-corrected C-indices of the models for AE were 0.67 [95% confidence interval (CI) 0.62-0.72] and 0.59 (95% CI 0.48-0.71), respectively. For nf/fAEs, ventricular arrhythmia severity before β-blocker initiation was a fourth independent predictor, and C-indices of the models in the derivation and validation cohorts were 0.74 (95% CI 0.68-0.80) and 0.60 (95% CI 0.47-0.72), respectively. In the derivation cohort, calibration slopes were 1.00 (95% CI 0.59-1.41) for AE and 1.00 (95% CI 0.69-1.32) for nf/fAE. CONCLUSIONS: These externally validated risk prediction models using clinical parameters accurately distinguished CPVT patients on β-blocker monotherapy at low and high risk for future AEs while treated with β-blockers. These models provide guidance for implementation of clinical management therapies to prevent AEs in patients with CPVT.

Humans↗
PubMed2001

Prevalidation of a model for predicting acute neutropenia by colony forming unit granulocyte/macrophage (CFU-GM) assay.

This report describes an international prevalidation study conducted to optimise the Standard Operating Procedure (SOP) for detecting myelosuppressive agents by CFU-GM assay and to study a model for predicting (by means of this in vitro hematopoietic assay) the acute xenobiotic exposure levels that cause maximum tolerated decreases in absolute neutrophil counts (ANC). In the first phase of the study (Protocol Refinement), two SOPs were assessed, by using two cell culture media (Test A, containing GM-CSF; and Test B, containing G-CSF, GM-CSF, IL-3, IL-6 and SCF), and the two tests were applied to cells from both human (bone marrow and umbilical cord blood) and mouse (bone marrow) CFU-GM. In the second phase (Protocol Transfer), the SOPs were transferred to four laboratories to verify the linearity of the assay response and its interlaboratory reproducibility. After a further phase (Protocol Performance), dedicated to a training set of six anticancer drugs (adriamycin, flavopindol, morpholino-doxorubicin, pyrazoloacridine, taxol and topotecan), a model for predicting neutropenia was verified. Results showed that the assay is linear under SOP conditions, and that the in vitro endpoints used by the clinical prediction model of neutropenia are highly reproducible within and between laboratories. Valid tests represented 95% of all tests attempted. The 90% inhibitory concentration values (IC(90)) from Test A and Test B accurately predicted the human maximum tolerated dose (MTD) for five of six and for four of six myelosuppressive anticancer drugs, respectively, that were selected as prototype xenobiotics. As expected, both tests failed to accurately predict the human MTD of a drug that is a likely protoxicant. It is concluded that Test A offers significant cost advantages compared to Test B, without any loss of performance or predictive accuracy. On the basis of these results, we proposed a formal Phase II validation study using the Test A SOP for 16-18 additional xenobiotics that represent the spectrum of haematotoxic potential.

Acute Disease↗
PubMed2003

Optimal growth strategies of larval helminths in their intermediate hosts.

We consider optimal growth of larval stages in complex parasite life cycles where there is no constraint because of host immune responses. Our model predicts an individual's asymptotic size in its intermediate host, with and without competition from conspecific larvae. We match observed variations in larval growth patterns in pseudophyllid cestodes with theoretical predictions of our model. If survival of the host is vital for transmission, larvae should reduce asymptotic size as intensity increases, to avoid killing the host. The life history strategy (LHS) model predicts a size reduction <1/intensity, thus increasing the parasite burden on the host. We discuss whether body size of competing parasites is an evolved LHS or simply reflects resource constraints (RC) on growth fixed by the host, leading to a constant total burden with intensity. Growth under competition appears comparable with "the tragedy of the commons", much analysed in social sciences. Our LHS prediction suggests that evolution generates a solution that seems cooperative but is actually selfish.

Animals↗
PubMed2006

A multimarker model to predict outcome in tamoxifen-treated breast cancer patients.

PURPOSE: This study was designed to produce a model to predict outcome in tamoxifen-treated breast cancer patients based on clinicopathologic features and multiple molecular markers. EXPERIMENTAL DESIGN: This was a retrospective study of 324 stage I to III female breast cancer patients treated with tamoxifen for whom standard clinicopathologic data and tumor tissue microarrays were available. Nine molecular markers were studied by semiquantitative immunohistochemistry and/or fluorescence in situ hybridization. Cox proportional hazards analysis was used to determine the contributions of each variable to disease-specific and overall survival, and machine learning was used to produce a model to predict patient outcome. RESULTS: On a univariate basis, the following features were significantly associated with worse survival: high pathologic tumor or nodal class, histologic grade, epidermal growth factor receptor, ERBB2, MYC, or TP53; absent estrogen receptor (ER) or progesterone receptor; and low BCL2. CCND1 and CDKN1B did not reach statistical significance. On a multivariate basis, nodal class, ER, and MYC were statistically significant as independent factors for survival. However, the benefit of ER-positive status was moderated by BCL2, ERBB2, and progesterone receptor. BCL2 and TP53 also interacted as an independent risk factor. A kernel partial least squares polynomial model was developed with an area under the receiver operating characteristic curve of 0.90. CONCLUSIONS: Our data show the predictive value of BCL2, ERBB2, MYC, and TP53 in addition to the standard hormone receptors and clinicopathologic features, and they show the importance of conditional interpretation of certain molecular markers. Our multimarker predictive model performed significantly better than standard guidelines.

Aged↗
PubMed1989

Clinical assessment of a published model to predict aminoglycoside-induced nephrotoxicity.

During the past decade, several patient risk factors have been identified as contributing to the development of aminoglycoside nephrotoxicity. Sawyers et al. recently published a method for estimating the probability of aminoglycoside nephrotoxicity on an individual patient basis. The present work represents a refinement of previous publications and has not been tested with the common variations used in aminoglycoside dosing. The purpose of this study was to determine both the qualitative and quantitative value of this method in predicting aminoglycoside induced nephrotoxicity. Eighty-three patients (47 male, 36 female) meeting the inclusion criteria of Sawyers et al. were entered into the study. Patient risk factors (age, sex, initial 1-h postinfusion aminoglycoside serum level, initial calculated creatinine clearance, duration of therapy, and presence of liver disease) were entered into a logistic regression analysis to determine the individual patient's risk of developing nephrotoxicity. These calculated probability scores were then compared with the observed nephrotoxicity in specific groups within our patient sample to see how effectively the model quantitatively performed. Twelve patients (14.5%) developed nephrotoxicity. The model predicted only 5 of the 12 patients developing nephrotoxicity (sensitivity or true positive = 42%). In the nonnephrotoxic group, the model accurately predicted only 38 of 71 patients (specificity or true negative = 54%). These data suggest that the model may accurately quantitate the number of patients likely to develop nephrotoxicity from a specific group but is unable to discriminate specific patients at risk of developing aminoglycoside-induced nephrotoxicity.

Adolescent↗
PubMed1992

[Logistic model for predicting the prognosis in patients with intra-abdominal infection based on the APACHE II index].

BACKGROUND: The APACHE II method has been widely used to classify patients according to disease severity. The high mortality and the lack of reliable prognostic parameters justify the study and application of indexes of severity (IS) and prognostic indexes (PI) in patients with intraabdominal sepsis. METHODS: A prospective study is presented concerning 100 patients with intraabdominal sepsis in whom a prognostic index derived from the APACHE II method by means of a logistic regression model was applied. In this model the APACHE II score is used as the only independent variable with the aim of predicting the outcome (mortality or survival) at the time of hospital admission. RESULTS: The predictive values obtained, with a 70% probability of mortality taken as the cut-off point, were: sensitivity 100%, specificity 87.7% and total predictive capacity 91%. CONCLUSIONS: It was concluded that the logistic prediction model of prognosis shows a high correlation with patient outcome and the application of the APACHE II score is useful in patients with intraabdominal infection.

Abdomen↗
PubMed2002

A prognostic model for predicting the disappearance of left atrial thrombi among candidates for percutaneous transvenous mitral commissurotomy.

OBJECTIVES: We sought to develop a prognostic model to predict the disappearance of left atrial thrombi (LAT) among candidates for percutaneous transvenous mitral commissurotomy (PTMC). BACKGROUND: Complete LAT resolution can be achieved with oral anticoagulation, allowing a number of patients to safely undergo PTMC. METHODS: We randomly allocated 108 PTMC candidates with LAT into two subsets---one to derive the model and the other to validate it. The existence of LAT and its size were measured by transesophageal echocardiography. Patients were given oral anticoagulation and followed up for 6 to 34 months. There was a 62% disappearance rate of LAT. RESULTS: We developed the following model: P = 1/(1 + exponential [-8.1 + 1.8 NYHA + 0.7 area]), where NYHA = New York Heart Association functional class (from I to IV), and area = LAT area (in cm(2)). The model was well calibrated (goodness-of-fit test, p = 0.82) and well discriminated (area under the receiver-operating characteristics [ROC] curve = 0.92). Performance in the validating sample was equally good (area under the ROC curve = 0.94; goodness-of-fit test, p = 0.16). When a cut-off point of p > 0.7 was used to designate the LAT disappearance in the validating set, the model had a sensitivity, specificity and positive and negative predictive values of 93.3%, 79.2%, 84.9% and 90.5%, respectively. CONCLUSIONS: Combined clinical (NYHA functional class) and echocardiographic (LAT area) variables are predictive of the 34-month outcome of oral anticoagulation for LAT resolution among PTMC candidates. This simple and highly predictive model might be potentially useful for clinical assessment and proper management.

Adult↗
PubMed2000

A model for predicting surgical outcome in patients with advanced ovarian carcinoma using computed tomography.

BACKGROUND: A reliable model for predicting the outcome of primary cytoreductive surgery may be a useful tool in the clinical management of patients with advanced epithelial ovarian carcinoma. METHODS: Forty-one women with a preoperative computed tomographic (CT) scan of the abdomen and pelvis and a histologic diagnosis of Stage III or IV epithelial ovarian carcinoma following primary surgery performed by one of nine gynecologic oncologists were identified from tumor registry databases. All CT scans were analyzed retrospectively using a panel of 25 radiographic features without knowledge of the operative findings. Patient demographics, surgical findings and outcome, Gynecologic Oncology Group performance status, and pre-operative serum CA125 values were collected from patient medical records. Residual disease measuring < or = 1 cm in maximal diameter was considered an optimal surgical result. Sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were calculated for each radiographic feature and clinical characteristic. Based on statistical probability of each factor predicting cytoreductive outcome, 13 radiographic features, in addition to performance status, were selected for inclusion in the final model. Each parameter was assigned a numeric value based on the strength of statistical association, and a total Predictive Index score was tabulated for each patient. Receiver operating characteristic (ROC) curve analysis was used to assess the ability of the model to predict surgical outcome. Statistical significance was evaluated using the Fisher exact test. RESULTS: Twenty of 41 patients (48.8%) underwent optimal cytoreduction to </= 1 cm residual disease. CT features of peritoneal thickening, peritoneal implants (>/= 2 cm), bowel mesentery involvement (>/= 2 cm), suprarenal paraaortic lymph nodes (>/= 1 cm), omental extension (spleen, stomach, or lesser sac), and pelvic sidewall involvement and/or hydroureter were most strongly associated with surgical outcome. Using the Predictive Index scores, a receiver operating characteristic curve was generated with an area under the curve = 0. 969 +/- 0.023. In the final model, a Predictive Index score >/= 4 had the highest overall accuracy at 92.7% and identified patients undergoing suboptimal surgery with a sensitivity of 100% (21/21). The specificity, or ability to identify patients undergoing optimal surgery, was 85.0% (17/20). The PPV of a Predictive Index score >/= 4 was 87.5% (21/24), and the NPV was 100%. The ability of this model to correctly predict surgical outcome was statistically significant (P < 0.001). CONCLUSIONS: In this model, a Predictive Index score >/= 4 demonstrated high sensitivity, specificity, PPV, and NPV, and was highly accurate in identifying patients with advanced epithelial ovarian carcinoma unlikely to undergo optimal primary cytoreductive surgery. The Predictive Index model may have clinical utility in guiding the management of patients with ovarian carcinoma.

Adult↗
PubMed2026

Systematic review of machine learning approaches for predicting sickle cell crisis and mortality risk at the climate-health nexus.

BACKGROUND: Sickle cell anemia (SCA) is a severe genetic blood disorder characterized by recurrent vaso-occlusive crises and increased mortality, with the greatest burden occurring in low- and middle-income countries. Climatic and environmental conditions, including temperature variability, humidity, rainfall, air pollution, and seasonal changes, have been associated with disease exacerbation. However, the extent to which these factors have been incorporated into predictive models remains unclear. This study systematically reviews the application of machine learning (ML) models for predicting SCA crises and mortality in relation to climate and environmental factors. METHODOLOGY: The PRISMA guidelines were used, and 34 peer-reviewed studies published between 2005 and 2026 were analyzed to identify the climate variables, ML approaches employed, and predictive performance. The reviewed studies applied a range of ML techniques, including artificial neural networks, random forests, support vector machines, decision trees, logistic regression, and deep learning models. Temperature, humidity, rainfall, wind speed, air quality indicators, and seasonal patterns were the most frequently examined environmental variables. RESULTS: The findings indicate that most existing models rely predominantly on clinical and demographic data, with limited integration of climate information and inadequate representation of high-burden regions, especially Sub-Saharan Africa. Studies incorporating environmental variables reported improved predictive performance and highlighted the potential of climate-informed early warning systems for SCA management. CONCLUSION: The review recommends development of interdisciplinary, climate-aware ML frameworks, expansion of longitudinal environmental datasets, and increased research in underrepresented regions to support climate-resilient and patient-centered SCA care.

Humans↗
PubMed1994

Validation of the Gail et al. model for predicting individual breast cancer risk.

BACKGROUND: The Gail et al. model is considered the best available means for estimating an individual woman's risk of developing breast cancer. Such estimates are useful in decision making on the part of women, in designing prevention trials, and in targeting screening and prevention efforts. PURPOSE: Our purpose was to evaluate the ability of the model to accurately predict individual breast cancer risk, using a large population independent of the one from which the model was derived. METHODS: We compared the number of cancer cases predicted by the model to the actual number of cases observed in the Nurses' Health Study. The study population was 115,172 women who did not have breast cancer at the beginning of the study. Questionnaires were sent to participants every 2 years, seeking data on risk factors and diagnoses of breast cancer. Follow-up compliance was 95% over the 12-year study period. RESULTS: The model over-predicted absolute breast cancer risk by 33% (95% confidence interval [CI] = 28%-39%), with the overprediction more than twofold among premenopausal women (95% CI = 1.9-2.2), among women with extensive family history of breast cancer (95% CI = 1.1-3.9), and among women with age at first birth younger than 20 years (95% CI = 1.3-4.7). The correlation coefficient between observed and predicted risk was 0.67, indicating that the model is less than satisfactory for ranking individual levels of breast cancer risk. Overprediction occurred at all deciles of predicted risk. CONCLUSIONS: The model's performance is unsatisfactory for estimating breast cancer risk for individual women aged 25-61 years who do not participate in annual screening. Lower mammography screening rates in the Nurses' Health Study may account for some, but not all, of the discrepancy between observed and predicted cases. IMPLICATIONS: A recent modification of the model by the tamoxifen trial investigators is likely to have provided accurate power calculations. This modified form of the model should be useful for planning other large, population-based studies.

Adult↗
PubMed2004

Use of a human visual system model to predict observer performance with CRT vs LCD display of images.

This Project evaluated a human visual system model (JNDmetrix) based on just noticeable difference (JND) and frequency-channel vision-modeling principles to assess whether a Cathode ray tube (CRT) or a liquid crystal display (LCD) monochrome display monitor would yield better observer performance in radiographic interpretation. Key physical characteristics, such as veiling glare and modulation transfer function (MTF) of the CRT and LCD were measured. Regions of interest from mammographic images with masses of different contrast levels were shown once on each display to six radiologists using a counterbalanced presentation order. The images were analyzed using the JNDmetrix model. Performance as measured by receiver operating characteristic (ROC) analysis was significantly better overall on the LCD display (P = 0.0120). The JNDmetrix model predicted the result (P = 0.0046) and correlation between human and computer observers was high (r (2) (quadratic) = 0.997). The results suggest that observer performance with LCD displays is superior to CRT viewing, at least for on-axis viewing.

Breast Neoplasms↗
PubMed1984

A molecular mechanical model to predict the helix twist angles of B-DNA.

We present here a model for the prediction of helix twist angles in B-DNA, a model composed of a collection of torsional springs. Statistically averaged conformational energy calculations show that, for a specified basepair step, the basepair-basepair conformational energy is quadratically dependent on the helix twist angle, so the calculations provide the spring parameters for the basepair-basepair interactions. Torsional springs can also be used to model the effects of the backbone on the helix twist, and the parameters for those springs are derived by fitting the model to experimental data. The model predicts a macroscopic torsional stiffness and a longitudinal compressibility (Young's modulus) which are both in good agreement with experiment. One biological consequence of the model is examined, the sequence specificity of the Eco RI restriction endonuclease, and it is shown that the discriminatory power of the enzyme receives a substantial contribution from the energetic cost of torsional deformations of the DNA when wrong sequences are forced into the enzyme binding site.

Base Composition↗
PubMed1999

The effects of sex ratio on sexual competition in the European lobster.

During the breeding season an individual's access to mates may be affected by operational sex ratios, causing strong variation in mating success. We manipulated adult sex ratios of the European lobster, Homarus gammarus, to test the predictions of models that relate sexual competition to (1) the sex ratio, (2) the time that an individual is not available to mate and (3) 'collateral investment', whereby two males contribute to a single clutch. The model predictions proved to be relatively insensitive to collateral investment. Male-male competition predominated in the male-biased but not in the female-biased sex ratio. This matches the predictions of one model that incorporates an extended period of female receptivity because the time that a male was unavailable to mate was small compared to the time spent by females in cohabitation and parental care. Although females increased their competitiveness when males were in the minority, male competition remained high. The insensitivity of male-male competition to sex ratios may be due to an upper limit to the costs that males can afford when there is a serious risk of injury, preventing males from increasing their aggression when females are in short supply. Copyright 1999 The Association for the Study of Animal Behaviour.

Journal Article↗
PubMed1998

Facile purification of fibrinogen fragments using a computer-based model with general applicability to the generation of salt gradients.

We and others have recently shown that specific fragments of cross-linked fibrin affect cell behavior. In order to develop a facile method for the preparative scale purification of fibrin fragment D dimer, a simple gradient generating system for conventional chromatography was developed and validated, and methods of fibrin fragment D dimer purification were compared. The experimentally measured salt concentration/time relationship fell directly on the model-predicted line. Model-predicted changes in the reservoir volume and/or salt concentration in the limit buffer affected both the initial slope and the shape of the concentration/time relationship. This gradient generation method was used to separate the D domains of fibrin(ogen) from the amino terminal region E domain using anion-exchange chromatography. While the predicted salt gradient was achieved, a salt-dependent separation was found to be less optimal than that of a pH-dependent separation, as validated by Coomassie-stained SDS-PAGE and by immunoblotting. In conclusion, a facile, user-friendly, computer-based method to predict and generate salt gradients was written and validated by direct experimentation. While fibrinogen fragment purification was acceptable using this system, both separation and yields of fibrinogen and fibrin fragments were superior using a pH-based separation technique.

Chromatography, Gel↗
PubMed2003

Model uncertainty and choices made by modelers: lessons learned from the International Atomic Energy Agency model intercomparisons.

The treatment of uncertainties associated with modeling and risk assessment has recently attracted significant attention. The methodology and guidance for dealing with parameter uncertainty have been fairly well developed and quantitative tools such as Monte Carlo modeling are often recommended. However, the issue of model uncertainty is still rarely addressed in practical applications of risk assessment. The use of several alternative models to derive a range of model outputs or risks is one of a few available techniques. This article addresses the often-overlooked issue of what we call "modeler uncertainty," i.e., difference in problem formulation, model implementation, and parameter selection originating from subjective interpretation of the problem at hand. This study uses results from the Fruit Working Group, which was created under the International Atomic Energy Agency (IAEA) BIOMASS program (BIOsphere Modeling and ASSessment). Model-model and model-data intercomparisons reviewed in this study were conducted by the working group for a total of three different scenarios. The greatest uncertainty was found to result from modelers' interpretation of scenarios and approximations made by modelers. In scenarios that were unclear for modelers, the initial differences in model predictions were as high as seven orders of magnitude. Only after several meetings and discussions about specific assumptions did the differences in predictions by various models merge. Our study shows that parameter uncertainty (as evaluated by a probabilistic Monte Carlo assessment) may have contributed over one order of magnitude to the overall modeling uncertainty. The final model predictions ranged between one and three orders of magnitude, depending on the specific scenario. This study illustrates the importance of problem formulation and implementation of an analytic-deliberative process in risk characterization.

Journal Article↗
PubMed2026

Large-scale multiethnic electronic health record resource for diabetes complications research: the North West London Diabetes Cohort (NWLDC) - cohort profile.

PURPOSE: The North West London Diabetes Cohort is established to provide systematic characterisation of a large diabetes population as a foundation for complications research and prognostic modelling. Many predictive modelling studies neglect the essential descriptive characterisation of underlying cohorts, focusing narrowly on model accuracy. This cohort profile addresses this gap by comprehensively describing the demographic composition, clinical characteristics and complication incidence patterns. The notably diverse, multiethnic population enables examination of ethnic disparities and supports future development of reliable prognostic models and evidence-based prevention strategies for diabetes complications. PARTICIPANTS: At baseline, 337&#x2009;271&#x2009;patients with diabetes were identified. It includes 279&#x2009;067&#x2009;patients with type 2 diabetes, 17&#x2009;638 with type 1 diabetes, 33&#x2009;590 with gestational diabetes and 6916 with unspecified diabetes. The earliest diabetes diagnosis dates to January 1932, with data updated to 27 May 2025. FINDINGS TO DATE: This cohort profile describes baseline characteristics of patients with comprehensive data collected on demographics (age, sex, Deprivation Index, ethnicity), clinical measures (glycated haemoglobin, body mass index, blood pressure, lipids and estimated glomerular filtration rate) and 14 major diabetes complications tracked longitudinally. Key findings for patients with type 2 diabetes reveal diabetic retinopathy as the most common complication (74.6 per 1000 person-years), followed by hypertension (51.0) and kidney disease (31.4). Cumulative incidence analyses using the Aalen-Johansen estimator, which accounts for mortality as a competing risk, demonstrated significant ethnic disparities, with black, Asian, mixed and other ethnic groups showing elevated risk compared with white patients. Time-varying Cox models identified strong clustering between cardiovascular and renal complications, confirming a cardiometabolic-renal syndrome. Mental health conditions (depression and anxiety) were prevalent throughout the disease timeline, occurring both before and after diabetes diagnosis. FUTURE PLANS: This cohort will be used as a platform for developing and validating prognostic models for diabetes complications, enabling risk stratification and targeted interventions. Future work will incorporate medication data to refine diabetes type classification, examine the effectiveness of antidiabetic medications in preventing different complications and address demographic differences in prognostic model performance and prediction accuracy. To better characterise lifestyle, further interrogation of electronic health record data will examine recording of advice given, including dietary advice, referral to weight management schemes and presence of alcohol consumption codes.

Humans↗